This assignment is for ETC5521 Assignment 1 by Team Possum comprising of Samuel Lyubic and Yuheng.
Volcanoes date back billions of years ago with the first recorded eruption in 1650 BC in Santorini. They are spread out across the globe with many lying along the pacific coast lines along earths tectonic plates. Volcanoes are contributiors to more then 80% of the earths surface which they are able to do through their extreme explosive force and contents, which has led them to have mapped and shaped the the world as it is understood today (https://www.nationalgeographic.com/environment/natural-disasters/volcanoes/). With the force volcanoes are able to generate, they can bring about life however their destructive nature is also understood, and sometimes unfortunately underestimated (https://www.volcanodiscovery.com/volcanic_risk_zones.html#:~:text=For%20instance%2C%20High%20Risk%20Zones,where%20volcanic%20projectiles%20fall%20regularly.&text=With%20bigger%20eruptions%20producing%20heavy,such%20pyroclastic%20flows%20are%20channeled.) with many scientist over the course of history attributing significant changes to the globe and historic climate events to volcanoes, given the range of elements that they emit and as well as the ferocity and magnitude of eruptions (Zeilinski (2015)). As such, the primary question of this report is _“What is the global impact of volcanic eruptions?” in order to better understand the risk volcanoes pose and the ramifications and reverberations to the human population from their eruptions.
The data source is from The Smithsonian Institution, which has been constantly updating since 2013. The data is cleaned and made available for download on (https://github.com/rfordatascience/tidytuesday/blob/master/data/2020/2020-05-12/readme.md), so that we directly import the 5 data sets for this analysis. These 5 data sets are probably the most comprehensive data set around. They are volcano, eruptions, events, tree_rings and sulfur, allowing us to discover the influences among them.
The data source is from The Smithsonian Institution. The data is available, and cleaned, downloadable from tidytuesday github. Cleaning script is also supplied.
Data provided contains 5 linked data sets, each with looking at a particular aspect of volcano:
volcano: Provides information on 958 volcanoes such asvolcano: Provides location, geological and population information on 958 volcanoes with 26 variables such as
eruptions: Details eruptions occured since Holecene period (11,345 years ago) to present, comprising of 15 variables. Details include
events: any event that occur at the volcanoes documented with 10 variables.
eruptions, thus it seems as a condensed version of eruptions.tree_ring: Tree rings were used as a climate proxy. In study of effects on volcanoes on climate change, researchers matched effects of eruptions to tree ring records. The measurements are conducted yearly and no data is missing. Threre will be more discussion about tree_ring later.
sulfur
Primary: What is the global impact of volcanic eruptions?
Secondary:
What is range of risk with volano eruptions? Do volcanoes with different VEI have specific characteristics? What is relationship between tree rings and volcanis eruptions? Can volcano eruptions predict climate change? Which tectonic settings have higher or lower VEI and how this relates to the duration of eruption? What are the ideal settings for eruptions to take place?
Figure 4.1: Number of eruptions for Primary Volcano Types
There are 3 main types of Active Volcanoes namely Stratovolcano, Caldera and Shield. The rest have been labelled as others. Stratovolcanoes are by far the most active volcano.
Figure 4.2: Active Volcanoes and The Ring of Fire, and Tectonic Plates layourt across the globe
Figure 4.2 displays two maps, one showing the Active Volcanoes and the ring of fire and the other showing the layour of tectonic plate boundaries.
We found that most of the volcanoes lie along the Pacific Ocean tracing the boundaries of tectonic plates, this is known as the ‘Ring of Fire’ (Society 2019). This path is approximately 40,000km long and holds 75% of the World’s Volcanoes with 90% of them being active. The abundance of volcanoes are explained by significant tectonic plates in the area, as volcanoes are formed through the cracking in the ground created by converging or diverging tectonic plate boundaries.
Figure 4.3: The Ring of Fire
Figure 4.3 displays the Ring Of Fire, which demonstrates the number of volcanoes running along the boundaries. Relatively speaking, the volcanoes here have more eruptions.
Figure 4.4: Population within 5, 10, 30, 100km from Volcano
Figure 4.4 displays the size of volcanoes as represented by the triangles and their proximity to the nearest population, broken down into 5km, 10km, 30km and 100km.
The impact of a volcano can vary in their destructive nature, creating both immediate and long term risks to the surrounding and global populations. The immediate risk to the surrounding populations come in the physical form of:
Size of a volcano can be an indicator of the potential damage that could come from an eruption given the positioning and greater velocity that can be prodcued from a larger volcano however, there is no way to create a blanket safe distance rule when it comes to their damage radius is the impact of a volcano uniqely dependent on each individual eruption. Generally, 5-10km range is relatively safe for residincy given the size of most eruptions however, as only an exceptional eruption would result in severe in that range, although these “exceptional eruptions” are not too uncommon for residents who decide to live in this proximity to setup imobile residencies. Volcanic Discovery%20Low%20Risk%20Zone&text=Typically%20you%20are%20more%20than,pyroclastic%20flows%20could%20be%20channeled.)
Figure 4.4 was inspired by Sil Aarts’ visualization where we use geom_polygon to create easy and intuitive visualizations.
Figure 4.5: World Map with countries experiencing most eruptions
Figure 4.5 displays the locations of volcanoes with a populations within a 5km radius, with the size of the population mapped to size of the point and orange scale indicating the number of volcano eruptions across the glove. with fill depending on the number of eruptions in each country and larger dot sizes represent the number of population in the vicinity.
Tree ring is an important indicator to reveal climate and temperature change. tree_rings dataset contains the values of two variables (n_tree and europe_temp_index) from year 0 to year 2000. n_tree is tree ring z-scores relative to year and z-score is a measure of variability from the mean). europe_temp_index is the average yearly temperature for Europe in Celsius relative to 1961 to 1990. We know that tree ring width reflects the growth conditions of the tree — the condition refers to nutrient conditions, precipitation and temperature during a growing season. Seiler, Houlié, and Cherubini (2017) suggest “the ring width may also be influenced by volcanic activity on Mount Etna and in other volcanic regions.” Seiler, Houlié, and Cherubini (2017) also suggest the pre-eruptive phase can only have begun when the trees had already ceased their seasonal growth.
tree_ring datasettree_ring dataset does not have the accurate tree ring z-score value near the volcanoes, and the temperature index is related to only Europe but not near the active volcanoes. The second limitation is because wildfires also affect the tree ring z-score, but we don’t have the wildfires records of Europe.
Due to tree_ring dataset only containing the temperature index and tree ring data in Europe, we filter the eruption events that occur in Europe only. However, most eruption events were in Russia, and partial Russian territory is belong to Europe. In other words, not all eruptions affect Europe temperature index. But it is still worthy to analyse the relationship between tree rings and volcanic eruption events by using the tree_ring dataset.
First, we draw a plot for n_tree and europe_temp_index. Fig 4.6 shows these the changes of two variables along time. The vertical dash lines indicate the eruption events. It seems that two lines generally fit each other. To ensure the correlation between these two variables, we run a correlation test. Table 4.1 shows the estimated correlation is 0.79 and p-value is less than 0.05. This is to say that we are 95% confident that two variables are strongly correlated. Thus, we can build a model for these two variables.
Figure 4.6: Tree ring z-score and temperature index
| estimate | statistic | p.value | parameter | conf.low | conf.high | method | alternative |
|---|---|---|---|---|---|---|---|
| 0.7931274 | 58.20812 | 0 | 1998 | 0.7762827 | 0.8088402 | Pearson’s product-moment correlation | two.sided |
We build two linear models: \[\large{lm\_1} :\large{\widehat{\text{Europe temp index}}} = 0.1073 + 0.4941 * \text{Tree ring z-score}\]
\[\large{lm\_2} :\large{\widehat{\text{Europe temp index}}} = 0.10765 + 0.49421 * \text{Tree ring z-score} - 0.02598 * \text{Eruption}\]
In lm_2, for eruption, 1 means there was an eruption while 0 means none.
Table 4.2 shows the goodness of two models. lm_2 is slightly better than lm_1 because its \(r^2\) is slightly greater than lm_1’s \(r^2\). In conclusion, both of them are not good because of low r_square. For example, lm_1 has \(r^2\) of 0.6287, which means only 62.87% of variation explained by the model. The reason might be an eruption has its radius of influence, and most of the eruptions were in Russia, thus, they did not have big influence in Europe temperature index. In addition, even there was a big eruption, it can hardly affect the whole Europe or the yearly average temperature index.
To visulising the fittness of lm_2, we plot the residual value based on lm_2. Residual value refers to the difference between observed and predicted value. If a model is a good model, the residual value would close to y intercept. Fig @ref(fig:lm_2_resid) shows the residual value spread, and variation of residual value is increasing while the temperature index is more far away from 0. This means, in some degree, lm_2 can predict the europe_temp_index under non-extreme value of n_tree.
| model | r_square |
|---|---|
| lm_1 (no eruption) | 0.6288653 |
| lm_2 (eruption) | 0.6287130 |
(#fig:lm_2_resid)Residual lm_2
The following section will be conducting analysis on volcano eruptions in order to assess how the elements of en eruption may impact the atmosphere and the resulting impact on climate.
<<<<<<< Updated upstreamAs discussed, a volcanic eruption can vary in VEI magnitude and range in the substances that are released however, a constant element that is emitted during eruption is Sulfur, which is subsequently also one of the most effective elements in cooling the climate. When sulfur is released into the stratosphere it combines with water to form sulfuric acid aerosols that produce a coating of small droplets in the upper stratosphere which act as a reflector to incoming solar radiation and result in a global coolng of the earths surface, with the aerosols able to stay in the statrosphere for up to 3 year, eventually falling back to earth. Furthermore, as discussed tree ring scores can be used to act as a proxy in order to understand the weather at the time thus by assessing the linearage of sulfur events and tree ring scores the impact of volcanic eruptions on climate change can be assessed (UCAR (2020)).
=======As discussed, a volcanic eruption can vary in VEI magnitude and range in the substances that are released however, a constant element that is emitted during eruption is Sulfur, which is subsequently also one of the most effective elements in cooling the climate. When sulfur is released into the stratosphere it combines with water to form sulfuric acid aerosols that produce a coating of small droplets in the upper stratosphere which act as a reflector to incoming solar radiation and result in a global coolng of the earths surface, with the aerosols able to stay in the statrosphere for up to 3 year, eventually falling back to earth. Furthermore, as discussed tree ring scores can be used to act as a proxy in order to understand the weather at the time thus by assessing the linearage of sulfur events and tree ring scores the impact of volcanic eruptions on climate change can be assessed [https://scied.ucar.edu/shortcontent/how-volcanoes-influence-climate].
>>>>>>> Stashed changesThe primary variables being used to assess the sulfur levels, in nannograms per gram, in the northern and southern hemisphere across the date range of 500-705 ce are:
Given the stable ice sheets as well as the earths rotation and locations of Greenland and Anatartica, the ice sheets abosrb and store the elements released during events in history and the NEEM and WDC team have been able to extract ice sheets that store records of atmospheric concentration of greenhouse gases, surface air temperature. Sulfur being one of these elements, thus the ice cores allow for a construction of a time series of the level of a sulfur at any given year thus allowing a reconstruction of when sulfuric events in the form of eruptions took place (NEEM (2020)) (WAIS (2020)).
The histroic time frame provides critical information on the history of eruptions and sulfuric events thus allowing to better understand what caused certain events in history, such as ice ages and prolonged periods of cooling, and long term effects of sulfuric depositions impact the earth in order to better understand the likey outcome on climate from an eruption and how to plan accordingly. [https://www.researchgate.net/publication/279965759_Timing_and_climate_forcing_of_volcanic_eruptions_for_the_past_2500_years]
Given the stable ice sheets as well as the earths rotation and locations of Greenland and Anatartica, the ice sheets abosrb and store the elements released during events in history and the NEEM and WDC team have been able to extract ice sheets that store records of atmospheric concentration of greenhouse gases, surface air temperature. Sulfur being one of these elements, thus the ice cores allow for a construction of a time series of the level of a sulfur at any given year thus allowing a reconstruction of when sulfuric events in the form of eruptions took place. [https://neem.dk/#:~:text=The%20North%20Greenland%20Eemian%20Ice,the%20previous%20interglacial%2C%20the%20Eemian.] [https://www.waisdivide.unh.edu/]
The histroic time frame provides critical information on the history of eruptions and sulfuric events thus allowing to better understand what caused certain events in history, such as ice ages and prolonged periods of cooling, and long term effects of sulfuric depositions impact the earth in order to better understand the likey outcome on climate from an eruption and how to plan accordingly. [https://www.researchgate.net/publication/279965759_Timing_and_climate_forcing_of_volcanic_eruptions_for_the_past_2500_years]
Figure 4.7: Time series of the sulfur events that have been recorded in Greenland and Antartica and the European weather index and Tree Ring Z-Score
Figure 4.7 displays three plots with year on the x axis and, for (A) and (B), historically recorded sulfur levels on the y axis while for (C) and (D), percentage change:
Looking at plot (A) and (B) the vertical lines indicate sulfuric events that the ice cores recorded and have been attributed to volcanic eruptions. Specifcally:
All these sulfur depositions have now been attributed to volcanic eruptions at the time (Sigl et al. (2015)) (Zeilinski (2015)). Asessing plot (C) and (D) from Figure 4.7, it is evident that there is negative change and tree growth z scores and climate index that occur after the volcanic eruptions. Specifically:
However, subsequent years after sulfur events show growth was again negative which indicates the lag and long term impact that sulfur depoistion events have on the stratosphere, as they may not always immediately show an impact however given the length of time the sufur can stay in the stratosphere it shows a long term effect taking place (Sigl et al. (2015)).
All these sulfur depositions have now been attributed to volcanic eruptions at the time [https://www.researchgate.net/publication/279965759_Timing_and_climate_forcing_of_volcanic_eruptions_for_the_past_2500_years] [https://www.smithsonianmag.com/science-nature/sixth-century-misery-tied-not-one-two-volcanic-eruptions-180955858/]. Asessing plot (C) and (D) from Figure 4.7, it is evident that there is negative change and tree growth z scores and climate index that occur after the volcanic eruptions. Specifically:
Year 536 saw one of the most significant sulfur depostions from an eruoption in the northern hemisphere with a negative change of -4800% from 0.7 to 3.29.
Interestingly, events in 574, 675, 627 and 690 are followed by growth in n tree ring size although still in the negative z score range.
However, subsequent years after sulfur events show growth was again negative which indicates the lag and long term impact that sulfur depoistion events have on the stratosphere, as they may not always immediately show an impact however given the length of time the sufur can stay in the stratosphere it shows a long term effect taking place [https://www.researchgate.net/publication/279965759_Timing_and_climate_forcing_of_volcanic_eruptions_for_the_past_2500_years].
Plot (D) displays the percentage change of the European temperature index, which displays a relatively negative percentage year on end change in temperature after a volcanic event.
>>>>>>> Stashed changesTo supplement the analysis conducted on Figure 4.7, a pearson correlation test was conducted on the tree scores and the yearly average NEEM and WDC sulfur level recordings.
| estimate | statistic | p.value | parameter | conf.low | conf.high | method | alternative |
|---|---|---|---|---|---|---|---|
| -0.3642468 | -5.586246 | 1e-07 | 204 | -0.4771875 | -0.2394694 | Pearson’s product-moment correlation | two.sided |
| estimate | statistic | p.value | parameter | conf.low | conf.high | method | alternative |
|---|---|---|---|---|---|---|---|
| -0.291919 | -4.359315 | 2.07e-05 | 204 | -0.4121722 | -0.1616692 | Pearson’s product-moment correlation | two.sided |
The results from Table 4.3 and Table 4.4 present a negative correlation between the tree ring sizes and the sulfur levels recorded by MEEN and WDC, with: - The correlation between tree rings and MEEN mean sulfur levels: -0.36. - The correlation between tree rings and WDC mean sulfur levels: -0.29.
Thus indicating that with increased sulfur levels tree ring size decreases and given tree rings act as a proxy for weather this leads to the notion that with event of a volcanic eruption taking and the subsequent release of the sulfur gases into the stratosphere there is likely to be both short term and long term climate impact with a global cooling period taking place.
VEI measures the explosiveness of volcanic eruptions which is determined by the volume of contents launched from the eruption. The contents is made up of pyroclastic flow, ash clouds, debrit and rocks with the eruption cloud height and volume being assessed as part of the measure. VEI is qualitatlively described using terms such ranging from “gentle” to “mega-colossal”. [Wikipedia].
Figure 4.8: Probability density function of eruptions in each VEI category
Theoretically, VEI ranges from 0 to infinity. In this dataset VEI has been recorded on a log10 scale, thereby each interval (increase of 1 in VEI) indicates an eruption 10x the magnitude. There only bee 40 eruptions over the last 132 million years that recorded a VEI-8 magnitude eruption and only in the last 10,000 years, only 10 eruptions have recorded a VEI-7.
Figure 4.8 presents the number of eruptions for the recorded VEI eruptions since 1812, with the number of eruptions on the x axis and the VEI rating on the y axis. The figure shows:
Figure 4.9: Frequency of Most Active Volcanoes in each VEI category
Figure 4.9 displays the volcanoes with the largest count of eruptions for each VEI rating presented in Figure 4.8. The frequency of eruptions are shown by the red tiles, with each tile representing an eruption, the year along the x axis with the name of the volcano and the corresponding VEI rating on the y axis.
The following section will be analysis the duration of eruptions for the range of tectonic settings.
Figure 4.11: Eruption duration across the range of tectonic settings
Figure 4.11 displays two figures, one for tectonic settings with long durations - greater then 365 days, and another for short durations - less then 30 days, with the count of the number of eruptions across the range of tectonic plates along the x axis and the tectonic setting on the y axis.
For long eruption durations:
Moreover, comparing the count figures, it is apparent that every number from long duration eruptions for all tectonic settings increase when it comes to short duration thus indicating that short duration eruptions occur more frequently than longer ones.
The following section will be analysing which how frequently each tectonic setting erupts, specifically assessing high frequency counts that involve two or more eruptions per year,
Figure 4.12: The high or low frequncies associated with the range of tectonic settings
Figure 4.12 displays the tectonic settings with the record of at least two or more volcano eruptions within one year with the count along the x axis and the tectonic setting on the y axis. Out of 10 tectonic settings recorded (NAs excluded), only 6 of them are listed under this condition. Tectonic settings with both long and short duration eruptions have all been listed here in regards with high frequency.
Grouped by 10 tectonic settings (NAs excluded), volcano eruptions on Rift zone / Intermediate crust (15-25 km) have the highest average VEI of 2.667, whereas those located on Intraplate / Oceanic crust ( <15 km) obtain the lowest average VEI of 0.923. Furthermore, as shown by Figure 4.11, despite Subduction zone / Continental crust ( >25km) having most longer ( >365 days) and shorter duration ( <30 days) eruptions occur as well as having the most higher frequency ( ≥ 2 eruptions within 1 year) of eruptions, it does not incur a high average VEI. It instead lies in the middle range among the 10 tectonic settings, with the average VEI of 1.982.
Overall, with the highest VEI average (2.667), neither long duration eruptions ( >365 days), nor high frequency (≥ 2 eruptions in 1 year) eruptions are found in Rift zone / Intermediate crust (15-25 km), while only one eruption of short duration ( <30 days) occurs in this setting. On the opposite, with the lowest ranking for average VEI (0.923), volcanoes on the Intraplate / Oceanic crust ( <15 km) don’t tend to erupt with long duration ( >365 days) where it’s long duration eruption only holds 4.575%, compared with other tectonic settings. Nevertheless, it has a greater percentage for both short duration (5.033%) and high frequency (9.677%) eruptions.
| tectonic_settings | Avg vei |
|---|---|
| Rift zone / Intermediate crust (15-25 km) | 2.667 |
| Intraplate / Continental crust (>25 km) | 2.250 |
| Subduction zone / Intermediate crust (15-25 km) | 2.195 |
| Rift zone / Oceanic crust (< 15 km) | 2.132 |
| Intraplate / Intermediate crust (15-25 km) | 2.091 |
| Subduction zone / Continental crust (>25 km) | 1.982 |
| Subduction zone / Oceanic crust (< 15 km) | 1.796 |
| Rift zone / Continental crust (>25 km) | 1.714 |
| Subduction zone / Crustal thickness unknown | 1.623 |
| Intraplate / Oceanic crust (< 15 km) | 0.923 |
It is evident from the analysis conducted that volcanic eruptions have a signifcant impact both to local populations and globally. From the analysis conducted it is evident that volcanoes can be unpredictable in VEI magnitude which leads to the notion that there is no blanket range that can followed to the letter that would be deemed safe or unsafe however, for the most part between 5km and 10km range would be considered safe in a mobile home setup. Despite the unpredictabiltiy on a case by case basis of each eruption it would be safest to not populate areas around volcanoes with the tectonic settings of Subduction zone / Continental crust ( >25km) Rift zone / Oceanic crust ( <15km) as these settings are ideal for eruptions and produce a larger number of eruptions, while it would be considered safest to populate areas along Intraplate / Intermediate crust (15-25km) as they tend to be associated with volcanoes that errupt the least frequently with lowest VEI.
<<<<<<< Updated upstreamFurthremore, by assessing the correlation relationship between tree rings and volcano eruptions, it clear that a strong correlation exits between tree ring z-score and temperature index. But, due to the limitations of tree_ring dataset, we cannot build a good model for temperature index. The model (lm_2) can only predict the temperature index well except for extreme values of weather index. To build a better model, we need more accurate temperature index around volcanoes. Following on, Figure 4.7 displays the percentage change in european weather index and tree ring size with the differing levels of sulfur recorded with a negative correlation exsiting between tree ring score sizes and both yearly mean MEEN and WDC sulfur levels inficating that after a volcanic eruption there is likely to be a subsequent period of cooling and climate change, showing the critical impact that volcano eruptions have both locally and globally.
Furthremore, by assessing the correlation relationship between tree rings and volcano eruptions, it clear that a strong correlation exits between tree ring z-score and temperature index. But, due to the limitations of tree_ring dataset, we cannot build a good model for temperature index. The model (lm_2) can only predict the temperature index well except for extreme values of temperature index. To build a better model, we need more accurate temperature index around volcanoes. Following on, Figure 4.7 displays the percentage change in european weather index and tree ring size with the differing levels of sulfur recorded with a negative correlation exsiting between tree ring score sizes and both yearly mean MEEN and WDC sulfur levels inficating that after a volcanic eruption there is likely to be a subsequent period of cooling and climate change, showing the critical impact that volcano eruptions have both locally and globally.
Data source is from Institution (n.d.).
In order to map the tectonic boundaries the tectonic plates were split at the prime meridian (x-intercept where long = 0) with the assistiance of Z.Lin at StackOverflow. The datasets were joined with the world map datasets provided by ggplot to find the continent and country coordinates as well as dataset from Kaggle to obtain coordinates to plot the tectonic plates polygon.
These packages are used to produce this report:
tidyverse (Wickham et al. 2019), lubridate (Grolemund and Wickham 2011), broom (Robinson, Hayes, and Couch 2020), leaflet (Cheng, Karambelkar, and Xie 2019), ggmap (Kahle and Wickham 2013), mapview(Appelhans et al. 2020), viridis (Garnier 2018), rgdal (Bivand, Keitt, and Rowlingson 2020), kableExtra (Zhu 2020), gridExtra (Auguie 2017), readr (Wickham, Hester, and Francois 2018), knitr (Xie 2014), sf (Pebesma 2018), data.table (Dowle and Srinivasan 2020), ggthemes (Arnold 2019), maps (Richard A. Becker, Ray Brownrigg. Enhancements by Thomas P Minka, and Deckmyn. 2018), ggridges (Wilke 2020), rvest (Wickham 2020)
Appelhans, Tim, Florian Detsch, Christoph Reudenbach, and Stefan Woellauer. 2020. Mapview: Interactive Viewing of Spatial Data in R. https://CRAN.R-project.org/package=mapview.
Arnold, Jeffrey B. 2019. Ggthemes: Extra Themes, Scales and Geoms for ’Ggplot2’. https://CRAN.R-project.org/package=ggthemes.
Auguie, Baptiste. 2017. GridExtra: Miscellaneous Functions for "Grid" Graphics. https://CRAN.R-project.org/package=gridExtra.
Bivand, Roger, Tim Keitt, and Barry Rowlingson. 2020. Rgdal: Bindings for the ’Geospatial’ Data Abstraction Library. https://CRAN.R-project.org/package=rgdal.
Cheng, Joe, Bhaskar Karambelkar, and Yihui Xie. 2019. Leaflet: Create Interactive Web Maps with the Javascript ’Leaflet’ Library. https://CRAN.R-project.org/package=leaflet.
Dowle, Matt, and Arun Srinivasan. 2020. Data.table: Extension of ‘Data.frame‘. https://CRAN.R-project.org/package=data.table.
Garnier, Simon. 2018. Viridis: Default Color Maps from ’Matplotlib’. https://CRAN.R-project.org/package=viridis.
Grolemund, Garrett, and Hadley Wickham. 2011. “Dates and Times Made Easy with lubridate.” Journal of Statistical Software 40 (3): 1–25. http://www.jstatsoft.org/v40/i03/.
Institution, Smithsonian. n.d. “Global Volcanism Program.” Smithsonian Institution | Global Volcanism Program. https://volcano.si.edu/.
Kahle, David, and Hadley Wickham. 2013. “Ggmap: Spatial Visualization with Ggplot2.” The R Journal 5 (1): 144–61. https://journal.r-project.org/archive/2013-1/kahle-wickham.pdf.
NEEM. 2020. North Greeland Ice Drilling. https://neem.dk/#:~:text=The%20North%20Greenland%20Eemian%20Ice,the%20previous%20interglacial%2C%20the%20Eemian.
Pebesma, Edzer. 2018. “Simple Features for R: Standardized Support for Spatial Vector Data.” The R Journal 10 (1): 439–46. https://doi.org/10.32614/RJ-2018-009.
Richard A. Becker, Original S code by, Allan R. Wilks. R version by Ray Brownrigg. Enhancements by Thomas P Minka, and Alex Deckmyn. 2018. Maps: Draw Geographical Maps. https://CRAN.R-project.org/package=maps.
Robinson, David, Alex Hayes, and Simon Couch. 2020. Broom: Convert Statistical Objects into Tidy Tibbles. https://CRAN.R-project.org/package=broom.
Seiler, Ruedi, Nicolas Houlié, and Paolo Cherubini. 2017. “Tree-Ring Width Reveals the Preparation of the 1974 Mt. Etna Eruption.” Scientific Reports 7. Nature Publishing Group: 44019.
Sigl, Michael, Mai Winstrup, Joseph Mcconnell, K. Welten, Gill Plunkett, Francis Ludlow, Ulf Büntgen, et al. 2015. “Timing and Climate Forcing of Volcanic Eruptions for the Past 2,500 Years.” Nature 523 (July). https://doi.org/10.1038/nature14565.
Society, National Geographic. 2019. “Ring of Fire.” National Geographic, April. https://www.nationalgeographic.org/encyclopedia/ring-fire/.
UCAR. 2020. How Volcanoes Influence Climate. How Volcanoes Influence Climate.
WAIS. 2020. WAIS Divide Ice Core. https://www.waisdivide.unh.edu/.
Wickham, Hadley. 2020. Rvest: Easily Harvest (Scrape) Web Pages. https://CRAN.R-project.org/package=rvest.
Wickham, Hadley, Mara Averick, Jennifer Bryan, Winston Chang, Lucy D’Agostino McGowan, Romain François, Garrett Grolemund, et al. 2019. “Welcome to the tidyverse.” Journal of Open Source Software 4 (43): 1686. https://doi.org/10.21105/joss.01686.
Wickham, Hadley, Jim Hester, and Romain Francois. 2018. Readr: Read Rectangular Text Data. https://CRAN.R-project.org/package=readr.
Wilke, Claus O. 2020. Ggridges: Ridgeline Plots in ’Ggplot2’. https://CRAN.R-project.org/package=ggridges.
Xie, Yihui. 2014. “Knitr: A Comprehensive Tool for Reproducible Research in R.” In Implementing Reproducible Computational Research, edited by Victoria Stodden, Friedrich Leisch, and Roger D. Peng. Chapman; Hall/CRC. http://www.crcpress.com/product/isbn/9781466561595.
Zeilinski, S. 2015. Sixth-Century Misery Tied to Not One, but Two, Volcanic Eruptions. https://www.smithsonianmag.com/science-nature/sixth-century-misery-tied-not-one-two-volcanic-eruptions-180955858/.
Zhu, Hao. 2020. KableExtra: Construct Complex Table with ’Kable’ and Pipe Syntax. https://CRAN.R-project.org/package=kableExtra.